Update README.md

README.md

@@ -36,7 +36,7 @@ engineer.
 | **Built-in agent** | `POST /api/agent` and MCP tool `agent_analyze` | One natural-language call; the server's own LLM loop plans and runs the tools |
 | **Files** | `GET /files/{session_id}/{filename}` | Generated SWMR docx / audit-zip downloads |
 
-## Tools (18)
+## Tools (19)
 
 `upload_model` → `run_simulation` → then: `get_node_results`, `get_link_results`,
 `get_subcatchment_results`, `get_timeseries` (bounded series with authoritative
@@ -44,6 +44,8 @@ engineer.
 store incl. the complete tokenized INP), `get_table_catalog`,
 `calgary_screening`, `set_report_details` (site description, design objectives,
 methodology, project metadata — see the swmr-site-details skill),
+`set_report_configuration` (project-specific major routes, criteria,
+classifications, drawing inventory, applicable reports and checklist overrides),
 `preliminary_design_review` (deterministic findings register incl.
 RPT-### reconciliation findings), `get_reconciliation`, `run_scenario` (base never
 mutated; deterministic comparison), `attach_figure` (embed client-generated
@@ -64,8 +66,34 @@ audited tables and provenance package. Instead:
    illustrative*, and it is archived in the audit zip under `figures/` with a
    manifest. The report itself remains the server's verified artifact.
 
-Typical flow: **upload_model → run_simulation → analysis tools**, reusing the
-returned `session_id`. Sessions expire after 6 h of inactivity.
+Typical CoC flow: **upload_model → set_report_details →
+set_report_configuration → run_simulation → deterministic QA/QC and Calgary
+screening → generate_report**, reusing the returned `session_id`. Sessions
+expire after 6 h of inactivity.
+
+Generated Calgary-style SWMR drafts include a deterministic depth-velocity
+criteria figure immediately below Table 9. It plots the tabulated 2011
+Alberta/Calgary envelope, the modeled overland-route depth/velocity pairs, and
+peak flow by marker colour. The figure PNG and its criterion CSV are retained
+in the audit ZIP. The report discloses straight-line interpolation and the need
+to verify current and project-specific requirements.
+
+The report also generates Figure 3-1, an automated SWMM model schematic, from
+the tokenized INP. It uses `[COORDINATES]` and `[VERTICES]` when available,
+falls back to a stable topology layout when coordinates are incomplete, shows
+subcatchment runoff routing and hydraulic-link direction, and archives both the
+PNG and a generation manifest. It is a topology aid and remains subject to
+drawing-to-model reconciliation.
+
+### Legacy zero-value solver options
+
+For dynamic-wave models, explicit zero values for `MAX_TRIALS`,
+`HEAD_TOLERANCE`, and `MIN_SURFAREA` are treated as legacy/default sentinels.
+The uploaded INP remains immutable. The server creates and runs a derivative
+execution copy with unit-aware EPA SWMM defaults, and records every
+substitution plus the SHA-256 hashes of both files. Negative or non-numeric
+values remain blocking configuration errors. The report audit ZIP retains both
+the original and derivative INP files whenever normalization occurs.
 
 ## Connecting from each platform
 
@@ -183,4 +211,4 @@ python -m venv /opt/swmm-venv && /opt/swmm-venv/bin/pip install -r worker-requir
 SWMM_WORKER_PYTHON=/opt/swmm-venv/bin/python uvicorn server:app --port 7860
 ```
 
-Free and non-commercial, for the water-engineering community.
+Free and non-commercial, for the water-engineering community.